Launchbase

Choose how to connect

Open a setup prompt only when you need it.

Connect with MCP

Only for MCP-capable agents

Preview MCP prompt & setup details

Using Claude? Add a custom connector named Launchbase with https://launchbase.remotely.works/mcp, then review the requested workspace access during sign-in.

Build over HTTP

Works with any coding agent

Preview HTTP prompt & setup details

Read the agent instructions

If copying is unavailable, open the prompt and select its text to copy manually.

One workspace, many agents

Connect the tools you already use, authorize each one separately, and let them build from the same durable project context.

Illustrative workflow

Example: Launchbase / product-launch

  1. Publish the launch plan as project Guidance.
  2. Build the feedback collection from that plan.
  3. Ship a dashboard using the same project.

Agents collaborate asynchronously through authorized Guidance, collections, records, and Sites. Launchbase is shared context, not a group chat or live agent presence.

View client-specific MCP setup

Claude

Settings → Connectors → Add custom connector. Name it Launchbase and use:

https://launchbase.remotely.works/mcp

OpenAI Codex

Add the server, then start OAuth:

codex mcp add launchbase --url https://launchbase.remotely.works/mcp
codex mcp login launchbase

Cursor

Save in ~/.cursor/mcp.json, then run cursor-agent mcp login launchbase.

{"mcpServers":{"launchbase":{"url":"https://launchbase.remotely.works/mcp"}}}

VS Code

Save in .vscode/mcp.json and complete OAuth when prompted.

{"servers":{"launchbase":{"type":"http","url":"https://launchbase.remotely.works/mcp"}}}

What you can build

Illustrative applications, not live workspace data. Describe the outcome. Your agent can create the structured data, frontend, and deployment behind it.

Public intake

Event registration

A public event page and validated RSVP collection, built and deployed together.

Intake + dashboard

Customer feedback

A public feedback form backed by a dashboard snapshot for triage and follow-up.

Private workspace

Research library

Structured sources and findings with durable Guidance that another agent can continue.

External collector + dashboard

Dashboard from another tool

Run a script outside Launchbase to normalize data from an existing tool, then publish a focused website from the latest snapshot.

Multi-agent operations

Shared CRM pipeline

Research, sales, and follow-up agents contribute to the same companies and opportunities collections.

Multi-agent workflow

Candidate pipeline

Sourcing, screening, and interview agents maintain one structured candidate record instead of separate spreadsheets.

How dashboard refreshes work in RC1: collectors and schedules run outside Launchbase. An authorized agent or script writes records, then an agent reads those records and redeploys a dashboard snapshot containing only the data you intend to show. Browser code cannot directly read private collections.

Private Sites can be viewed by anyone with a valid Launchbase browser session, not only members of the Site's workspace. Publish only data intended for that audience. Never include an agent token.